SciFIBench

What is SciFIBench?

SciFIBench is a scientific figure interpretation benchmark for LMMs, consisting of 2000 questions split between two tasks across 8 categories.

Released

2024-05-14

Modality

multimodal

Evaluates

多模态, Multimodal, 知识, Knowledge, NeurIPS 2024, 多模态模型, VLM, 跨模态推理, Cross-modal Reasoning, 科学推理, Scientific Reasoning, 不支持, Unsupported

Openness

restricted

Importer

OpenCompass Hub

Reported scores

0

Source provenance

• Original evidence https://hub.opencompass.org.cn/dataset-detail/SciFIBench

SciFIBench paper, code and dataset

• Paper https://arxiv.org/abs/2405.08807
• Code https://github.com/jonathan-roberts1/SciFIBench
• Dataset https://huggingface.co/datasets/jonathan-roberts1/SciFIBench

Which sources cite SciFIBench?

• SciFIBench

Related 多模态, Multimodal, 知识, Knowledge, NeurIPS 2024, 多模态模型, VLM, 跨模态推理, Cross-modal Reasoning, 科学推理, Scientific Reasoning, 不支持, Unsupported benchmarks

• 3DCodeBench
• A-Bench
• A-OKVQA
• AbsPyramid
• ActionAtlas
• ActiveFly-Bench
• Ada-LEval
• AECBench